Simulation Is Not Experiment
A simulation is a controlled argument from assumptions; an experiment is a question put to nature. Kronos values both and never mistakes one for the other.
What each one can and cannot do
A simulation computes what should happen if a set of physical assumptions and models are correct. A well-built simulation can be rigorous, reproducible, and predictive. What it cannot do is discover that one of its assumptions is wrong about the real world. Only an experiment does that. Kronos designs digitally first, then tests against hardware — and keeps the two clearly separated.
The value and the limit
- Simulation lets us explore a design space cheaply and safely before committing to steel
- It exposes sensitivities — how an output moves when an assumption moves
- It cannot certify that the modelled physics matches the built machine
- So a simulated Q, TBR, or field is a prediction awaiting a verdict
How we keep it honest
Our simulation outputs are published with the code and data to reproduce them, so a reader can check the argument. They are labelled as design points. And they are paired with pre-registered predictions, so that when hardware arrives, the simulation is scored, not quietly adjusted to match.
The strongest thing we can say about a simulation today is that it is reproducible and self-consistent. The strongest thing we cannot yet say is that it is right.